Remove Middle Dot (U+00B7) from URL (for Catalan only?)

[xavivars]

Currently, remove_accents() converts all characters to an ASCII equivalent so it looks "nice" as a URLs without the need of escaping characters (and, thus, showing % as part of the links).

However, the middle dot (U+00B7) is not removed. Middle dot is used in Catalan between two L (like this l·l).

Quoting from wikipedia:

The flown dot (Catalan: punt volat) is used in Catalan between two Ls in cases where each belongs to a separate syllable, for example cel·la, "cell". This distinguishes such "geminate Ls" (ela geminada), which are pronounced [ɫː], from "double L" (doble ela), which are written without the flown dot and are pronounced [ʎ].

On top of non being consistent (all other Catalan diacritics are removed), not removing this character has some side-effects, because there are some URL libraries that don't take it into account (like the one Twitter uses: see ​https://twitter.com/VilaWeb/status/738348674137399296).

My proposal is to remove that char when it appears between two l.

[xavivars]

Removes middle dot when Catalan is set as a language

[xavivars]

Removes middle dot for all languages

[ocean90]

@xavivars Thanks for your patches. The replacement should only be done for Catalan. Removing the dots can maybe handled by sanitize_title_with_dashes().

Can you make sure that the patches are relative to the root directory? And there should be a unit test for this change in /tests/phpunit/tests/formatting/RemoveAccents.php.

[xavivars]

@ocean90: should the patches be relative to the root directory of which repo? I've found contradictory information (sometimes pointing to develop.svn.wordpress.org and some other times pointing to core.svn).

I'll also add unit tests for that.

[swissspidy]

develop.svn.wordpress.org (or develop.git.wordpress.org) would be the correct repository for patches.

[xavivars]

Formatting.php patch

[xavivars]

RemoveAccents.php test patch

[xavivars]

@ocean90: I don't think I agree the removal of those dots should be done at sanitize_title_with_dashes. The middot it affects how the L are pronounced, and in fact, the first case was already covered in the same remove_accents method (I've removed it from the new formating.php.patch). However, if you think those changes belong better to sanitize_title_with_dashes, I'm open to discuss about that.

[SergeyBiryukov]

37086.patch​ combines the patch and the test and also updates the docs.

I think it's fine to handle this in remove_accents().

[xavivars]

Is there anything pending for this ticket to be closed that I can help with?

[ocean90]

In 37853:

I18N: Add support for the Catalan flown dot in remove_accents().

Props xavivars, SergeyBiryukov.
Fixes #37086.